基于模糊技术的非线性系统故障诊断与容错控制研究

Due to the continuously increasing system automation, integration and complexity degrees, industrial processes are typically nonlinear. Associated with increasing demands on safety and reliability for such processes, this project is devoted to investigate fuzzy technique based fault detection (FD) and fault-tolerant control (FTC) for general nonlinear systems. To this end, the parameterization forms for observer-based fault FD systems and controllers are studied first with the aid of nonlinear factorization technique. Based on the parameterization results, the architecture for the integrated design and optimization of nonlinear FD and FTC systems is established. The proposed architecture provides an integrated solution that has advantages to make the plant maintenance, repair and operations easier to handle. In addition, the existence conditions for different types of nonlinear FD systems are studied as well as the associated threshold settings. Then, different types of FD scheme are developed by dealing with the proposed conditions with the help of Takagi-Sugeno (T-S) fuzzy dynamic modeling technique. Based on the information provided by the fault diagnosis unit and the parameterization form of all the stabilizing controllers, the residual-driven dynamic controllers are proposed to compensate the influence of the faults without adjusting the current controller structure. .Moreover, considering that the uncertainties and unmodelled dynamics generally exist in industrial processes, this project is dedicated to investigate the data-driven identification and on-line optimization of nonlinear FD systems by adopting fuzzy technique as a solution tool. Then, the on-line optimization approach for the controllers and the fault-tolerant controllers of nonlinear processes are addressed with data-driven methods. It is worth mentioning that the research work in this project will contribute the study on the process monitoring and fault-tolerant control for nonlinear industrial processes.

现代工业系统结构越来越复杂，多呈现高度非线性特性。为保障此类系统的安全运行，本项目主要利用模糊技术研究一般非线性系统故障诊断与容错控制的设计方法。首先，借助非线性互质分解技术，研究一般非线性系统的故障诊断观测器与控制器的参数化形式，并进一步建立故障诊断与容错控制的集成设计与优化框架。其次，利用模糊动态建模技术逼近一般的非线性系统，并在此基础上研究不同类型的故障诊断观测器的设计，并根据故障诊断单元提供的信息和控制器参数化的结果，在不改变现有控制器结构的情况下，通过设计残差驱动的动态控制器来实现容错控制。再次，考虑到工业系统中普遍存在的不确定性和无法精确建模的问题，研究基于数据的非线性系统的故障诊断观测器的直接辨识与在线优化。最后，利用模糊技术研究基于数据的非线性系统的控制器在线优化和容错控制问题。本项目的研究成果为工业系统的实时监测与容错控制策略的制定提供有效的设计方法和理论依据。

为保障复杂工业系统的安全运行，本项目利用模糊与算子技术研究了一般非线性系统故障诊断与容错控制的设计方法。主要研究内容包括：借助非线性互质分解技术，研究了一般非线性系统的故障诊断观测器与控制器的参数化形式，并进一步建立故障诊断与容错控制的集成设计与优化框架；利用模糊动态建模技术逼近一般的非线性系统，并在此基础上研究了不同类型的故障诊断观测器的设计，并根据故障诊断单元提供的信息和控制器参数化的结果，在不改变现有控制器结构的情况下，通过设计残差驱动的动态控制器来实现了容错控制；考虑到工业系统中普遍存在的不确定性和无法精确建模的问题，研究了基于数据的非线性系统的故障诊断观测器的直接辨识与在线优化；研究了数据驱动的复杂工业系统控制器在线优化和容错控制问题。本项目的研究成果为工业系统的实时监测与容错控制策略的制定提供了有效的设计方法和理论依据。